Re: [PATCH v3 2/4] usb: aspeed-vhub: fix remote wakeup failure in iKVM use case

[Benjamin Herrenschmidt <benh@kernel.crashing.org>]

On Wed, 2021-12-08 at 18:05 +0800, Neal Liu wrote:
> Signaling remote wakeup if an emulated USB device has any activity
> if the device is allowed by host.
> 
> Signed-off-by: Neal Liu <neal_liu@aspeedtech.com>

I still think it should fundamentally be the device making that
decision, but since they don't, this is an acceptable workaround, but
please, don't write the MMIO on every EP queue. Either keep track of
the bus being suspended, or turn on the AUTO bit in HW when wakeup_en
is set.

Cheers,
Ben.

> ---
>  drivers/usb/gadget/udc/aspeed-vhub/epn.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/drivers/usb/gadget/udc/aspeed-vhub/epn.c
> b/drivers/usb/gadget/udc/aspeed-vhub/epn.c
> index 917892ca8753..ccc239b5cc17 100644
> --- a/drivers/usb/gadget/udc/aspeed-vhub/epn.c
> +++ b/drivers/usb/gadget/udc/aspeed-vhub/epn.c
> @@ -381,6 +381,11 @@ static int ast_vhub_epn_queue(struct usb_ep*
> u_ep, struct usb_request *u_req,
>  	} else
>  		u_req->dma = 0;
>  
> +	if (ep->dev->wakeup_en) {
> +		EPVDBG(ep, "Wakeup host first\n");
> +		ast_vhub_hub_wake_all(vhub);
> +	}
> +
>  	EPVDBG(ep, "enqueue req @%p\n", req);
>  	EPVDBG(ep, " l=%d dma=0x%x zero=%d noshort=%d noirq=%d
> is_in=%d\n",
>  	       u_req->length, (u32)u_req->dma, u_req->zero,